Run FFmpeg on Vercel
Offload the command and skip Vercel's 250 MB bundle cap.
Serverless FFmpeg
An FFmpeg API runs FFmpeg commands on cloud servers instead of your own machine. Rendobar's FFmpeg API runs the exact command you tested locally and returns the output URL on a webhook, from any runtime, including ones that can't spawn FFmpeg. Jobs run up to 1 hour on the free credits and 9 hours on Pro and Studio, billed per second.
$5 in free credits. No credit card required.
Run FFmpeg where you can't ship FFmpeg
Usage
import { createClient } from "@rendobar/sdk";
const rb = createClient({ apiKey: process.env.RENDOBAR_API_KEY });
const job = await rb.jobs.create({ type: "ffmpeg", params: { command: "ffmpeg -i https://cdn.rendobar.com/assets/examples/sample.mp4 -c:v libx264 -crf 23 out.mp4", },});
const result = await rb.jobs.wait(job.id);console.log(result.output); // { data, file, files, expiresAt }Simulated run with an example asset. Run real jobs in the playground.
Serverless
Edge and serverless functions can't ship a system binary or spawn a subprocess, so FFmpeg won't run there. Even ffmpeg.wasm can't run inside a Cloudflare Worker. Offload the command to Rendobar and the same call works from any runtime.
Offload the command and skip Vercel's 250 MB bundle cap.
Process video from a Worker that can't spawn a binary.
Trigger jobs from a Deno Edge Function over REST.
The binary is 140 MB of a 250 MB budget. Measured.
Read from the bucket and deliver back, with no access key.
Keep media in R2, where serving it costs nothing in egress.
20 common tasks, each command actually run.
24 filters, 72 variants, real encode times and costs.
60 seconds synchronous, and a hard 2 vCPU ceiling.
Netlify, Node, Deno, and any other runtime follow the same pattern.
Capabilities
Rendobar's FFmpeg API is FFmpeg as a service. Each operation below is one command sent to POST /jobs. No preset menu, no per-feature pricing. See the FFmpeg guide for the full reference.
These are the common ones. Anything else the latest FFmpeg can encode or decode works too. You run the real binary, not a limited subset. For commands that are already tested, the FFmpeg recipes and the FFmpeg filter reference show each one run on a real file, with output size, time and cost.
GPU acceleration
Rendobar's FFmpeg API runs your commands on CPU-powered or GPU-powered machines. The hardware encoders h264_nvenc, hevc_nvenc and av1_nvenc on NVIDIA L4, L40S or RTX PRO 6000 GPUs encode H.264, HEVC and AV1 in dedicated hardware. There are no GPU instances to provision and no drivers to install.
Pick the hardware with the optional compute param, or leave it on auto. Auto sends any command that already uses NVENC or CUDA to a GPU, so an existing ffmpeg command needs nothing new. Billing is per second of processing.
GPU acceleration is on the Pro and Studio plans. CPU and auto run on every plan. For archival masters or the lowest bitrates, software libx264 at a slow preset still wins on quality per bit, and you keep that option.
Integrations
Beyond a raw HTTP request: no-code automations, a typed SDK, and native MCP for AI agents.
Compare
Self-host FFmpeg and you own the servers, scaling, and security. Use a preset-based API and you lose raw control. Rendobar gives you both: every FFmpeg command, nothing to run.
| Compared on | Rendobar | Self-hosted FFmpeg | Preset-based APIs |
|---|---|---|---|
| Run any FFmpeg command | Yes | Yes | Presets only |
| Servers to run | None | You manage them | None |
| Time to first job | One POST | Hours to days | Minutes |
| Pay only for what runs | Per compute second | Servers bill 24/7 | Often GB-metered |
| Sandboxed per job | Yes, isolated | Your responsibility | Not exposed |
| Async webhooks + realtime | Built in | Build it yourself | Varies |
| Native MCP for AI agents | Yes | No | No |
| Zero egress on output | Yes | Depends on host | Often charged |
| FFmpeg upgrades + security | Managed | Your job | Managed |
| Free to start | $5, no card | Pay for infra | Card required |
Compared against the two common alternatives to a managed FFmpeg API. Comparing a specific provider? See Rendobar vs Rendi, vs Transloadit, vs Shotstack, or all comparisons. For eleven services side by side, with runtime caps and a measured job, read the best FFmpeg API in 2026.
FAQ
An FFmpeg API runs FFmpeg commands on a remote server instead of on your own machine. Rendobar's FFmpeg API does exactly that. You POST an FFmpeg command with input URLs, it runs them in a sandboxed cloud container, and returns the output URL on a webhook. You get the full power of FFmpeg without installing the binary or running servers.
Jobs are async by default. POST returns a job ID immediately. Rendobar sends a webhook to your URL on completion. You can also poll GET /jobs/{id} or subscribe via the realtime SDK.
Jobs are billed per second of compute. The median FFmpeg job we have measured cost $0.0025, and a long encode costs what it runs. Every account starts with $5 in free credits, no card required. Pro is $9/month for higher limits and a 9 hours per-job budget, and Studio is $29/month for 50 concurrent jobs. At low to medium volume this beats self-hosting FFmpeg, where idle servers bill 24/7 even when no job is running.
Free tier jobs run up to 1 hour. Pro and Studio raise that to 9 hours per job. That is a limit on processing time, not on how long your output can be. There is no cap on output length, so a long render is fine as long as it finishes inside the budget.
Not natively. Those runtimes can't ship the FFmpeg binary or spawn a subprocess, and Cloudflare Workers can't run ffmpeg.wasm because the core exceeds the 10 MB bundle limit. Instead you call Rendobar's FFmpeg API over HTTP from any of them, and the command runs in a sandboxed container. You pass public or presigned input URLs directly in the command. See the Vercel, Cloudflare, and Supabase guides for setup.
Each input can be up to 500 MB on the Free plan, 10 GB on Pro and 20 GB on Studio, whether you upload it or pass a URL. Contact us for higher limits.
Outputs stay available for 7 days on the free tier and 30 days on Pro and 90 on Studio. Download or copy them to your own storage before they expire.
You run the latest FFmpeg build, not a limited subset. That covers H.264, H.265 / HEVC, VP9, AV1, ProRes, AAC, Opus, and every other codec and filter the binary ships with.
Yes. Rendobar's FFmpeg API runs h264_nvenc, hevc_nvenc, and av1_nvenc on NVIDIA L4, L40S or RTX PRO 6000 GPUs, billed per second for the GPU the job ran on. Add a compute param to pick the hardware. It defaults to auto, so an NVENC or CUDA command routes to a GPU on its own. GPU acceleration is available on the Pro and Studio plans. CPU and auto work on every plan.
Yes, because the infrastructure ran. If a failure was caused by our infrastructure or a third-party outage, contact support for a discretionary refund.
MediaConvert and Coconut are preset-driven. You pick a preset, they encode. Rendobar runs the FFmpeg command you write. Anything FFmpeg can do, the API can do, including filter graphs that presets cannot express.
You write the FFmpeg command. We run it, from a single clip to your entire pipeline, and there are no servers for you to provision or patch.
Re-run on an ffmpeg job whose command reads a URL after -i works again, and a dashboard error now shows the API's reason instead of a generic message.
A URL inside a subtitles, movie or fontfile filter option is now downloaded with your other inputs, so asset URLs work there and a file is fetched once.
Send the FFmpeg command you would run locally. Rendobar runs it as written on FFmpeg 9.0 and returns every file it produces, with your own output names and ffmpeg's real errors.